First Look – Causata - SmartData Collective

Causata is a software company founded in early 2009 with a heritage from the online space – behavioral targeting on the web and database marketing. They are headquartered here in Silicon Valley with engineering in London and the team has experience across companies like Touch Clarity, Omniture, Interwoven, DoubleClick, RedHat, Skype, eBay and others.

Causata sees database marketing and online marketing merging. They see companies trying to consume large amounts of batch AND real-time data about customers. In particular they feel that web data – what you looked at on the web – is particularly important in targeting customers. Consumer-facing companies are under a lot of pressure to understand and act on all this data and do so quickly. Causata’s vision is to provide a solution that focuses on how to treat a specific customer by bringing together and analyzing these different data sources.

To do this, Causata assembles a complete view of the customer, enables access to predictive customer profiles in real-time and enables efficient real-time operational decisions across multiple touch points. In particular they focus on bringing web data like searches and pages view as well as data about mobile devices and combining it with offline data such as purchases, outbound contact history, demographic data etc. They:

Bring in a multitude of data sources, typically starting with web data and adding other sources over time.
Do real-time predictive modeling for online targeting
Provide data to 3^rd party modeling tools and consume the models that result
Bring this all together for real-time provisioning.

Key elements of the solution:

Multi channel with web at the center.
They bring data from devices and channels including the web and structure it around the customer. The solution manages uncertainty about user identity (shared computers for instance) and continually updates a “best customer view”. This includes retroactively mapping anonymous web data to registered customers and later purchases for instance. Causata allows you to access a customer profile using any customer identifier (e.g. cookie, email, customer ID, etc.).
Optimization based on business objectives
Even in online use cases the impact on overall business goals is what really matters. For instance improving conversation rates for people who are not ultimately profitable customers is unhelpful. Causata optimizes offers and behavior online based on the overall business objectives downstream of the online interaction itself.
Schema-free data management.
Causata uses event streams stored for each customer rather than a database schema. The event streams can be accessed in parallel, when building predictive analytics, as well as flexibly for a single customer. Summary data is not stored but calculated dynamically as and when needed. Their focus on customers simplifies this problem and makes for fast implementation.
Data structured for influencing outcomes.
Everything is designed around real-time identification of an optimal action for a specific customer. Events are kept in date/time order to help infer cause and effect – always an important point when building predictive analytic models.

They can generate scenario record datasets from their event streams. You can identify the “journeys” customers took to specific outcomes and filter the event streams down to these journeys. They align these journeys around the important interaction (like successful retention) and create records for each journey relative to the key interaction so you get all the events that led each customer to this destination in terms of time before that event.

This kind of dataset crosses all the different sources of data and combines them into a dataset with a strong time dimension – something ideally for predictive analytic modeling using a third party tool like SPSS, SAS or R. The models can be used by Causata either by scoring data and feeding that data back into Causata or by moving the model into Causata for execution.

In operations, Causata provides scores (propensity, lifetime value), rankings (customer decline, lead quality) and decisions (most relevant content, next best offer) as well as the complete set of customer profile information and delivers this through a set of standard APIs. This allows the extra information Causata is delivering to be injected into existing processes whether they be sending email or display a webpage or something else.

Causata provides various user interface elements for exploring and understanding the data and to support exporting data to third party tools. They also have their own automated modeling component that is focused on high volume online situations.

Causata provides both on-premise and hosted options but it is designed around a single customer installation. You can run it on premise or they can put dedicated systems up for you remotely.